TextWash 🫧

What are “AI fingerprints”?

Text generated by AI assistants tends to carry technical and stylistic markers that ordinary typed text doesn't:

• Invisible Unicode characters — zero-width spaces (U+200B), narrow no-break spaces (U+202F), soft hyphens, word joiners and BOMs. You can't see them, but detectors, plagiarism scanners, spam filters and diff tools can.
• Typographic tells — curly “smart” quotes, em-dashes (—), and real ellipsis characters (…) that almost nobody types by hand on a normal keyboard.
• Structural tells — Markdown residue pasted as plain text (**bold**, # headers, backtick code), emoji-decorated bullet lists, rhetorical fragments (“The result?”), and the “It's not X. It's Y.” contrast construction.
• Style clichés — words like delve, tapestry, and phrases like “it's important to note” that appear in AI text at wildly elevated rates.

What TextWash does

TextWash removes the technical fingerprints (invisible characters, smart punctuation) and flags the stylistic ones so you can rewrite them in your own voice. It never rewrites your words for you — the result is still your text, just clean. Works with pasted text or files: .txt, .md, .docx and .pdf, all processed locally.

Why trust it?

Because there's nothing to trust: the entire tool is a single page of open-source JavaScript that runs locally. No account, no server, no analytics on your text, no limits. Load the page, go offline, and it still works.